Rental yield is simply the monthly rent-to-house price ratio multiplied by 12. It therefore gives the annualized return one can expect from investment in housing above and beyond house price appreciation. An obvious corollary of the definition is that rental yields can fluctuate for two reasons: due to fluctuations in house prices or due to fluctuations in house rents. Following Matthew Klein, we examine data from Zillow. We look at the zip code level to understand what is going on in the US housing market.
In looking at US housing, it is important to keep in mind that it took a long time to recover from the boom-bust cycle of the mid-2000s. Investment in residential fixed assets (shown here as a percent of GDP) got coupled to the dealer leverage cycle in the early-2000s, started falling in 2006, and did not bottom out until 2011. From there it recovered very slowly until 2019. It is only now returning to historical mid-cycle levels.
Parenthetically, note that the FRED graph above shows that the business cycle is the housing cycle. The only recession not predicted by the turn of the residential investment cycle was 2001, which perhaps explains why it was so mild. In general, there is a clear dose-response relationship between the amplitude of the residential investment downcycle and the length and depth of US recessions.
But let’s look at rental yields. The Zillow data doesn’t go back beyond 2014. And the first two years of data seems to have integrity issues. So we shall look at what’s happened over the past five years, 2016-2021. The next graph shows the mean and median rental yield over zip codes.
Mean rental yield fell from 6.8 percent per annum in 2016 to 6.2 percent by 2019, where they stabilized for a year or so. Since the pandemic they have collapsed from 6.2 percent to 5.7 percent. Why? Is the collapse in rental yields due to falling rents or rising home prices?
At the aggregate level, rents have climbed steadily at 3.3 percent per annum. This is appreciably faster than inflation, suggesting that landlords have steadily gained at the expensive of renters.
Meanwhile, house prices have grown faster—especially since the pandemic. Over the past five years, house prices have appreciated at 4.9 percent per annum. Meanwhile, average rental yields have fallen by 3.1 percent per annum. These patterns suggest that the collapse in rental yields is largely due to a revival in house prices. We’ll see that this hypothesis holds up when we look at the data at a more granular level.
Since rental yield is simply the rent-to-price ratio, log yields are a linear function of log rents and log house prices. More precisely, log yields are the difference between log rents and log house prices. Because changes in log levels are growth rates, it greatly simplifies the interpretation to look at log changes in the three levels.
The next graph shows histograms for 5-year changes in log house prices, log house rents, and log rental yields. We can see that the mean growth rates are representative of the location of the distributions, which are approximately normal for prices and rents, and shifted in the manner suggested by the means.
The next set of graphs show scatter plots for changes in log yields against changes in log prices and log rents, respectively. They demonstrate that the collapse of rental yields is almost entirely due to house price appreciation. This is borne out by linear regression. Change in log prices explains 43.6 percent of the cross-sectional variation in change in log yields. Meanwhile, change in log rents explains only 0.3 percent of the variation, although the slope bears the right sign and is significant at the standard 5 percent level (P = 0.018).
Is there heterogeneity in the relationship between prices and yields? In particular, is house price appreciation concentrated in zip codes where house prices are high?
In order to answer this question, we stratify the sample by house price quartiles (ie, four equal buckets). The scatter plot reveals no significant difference in the relationship between house price appreciation and falling yields across house price quartiles.
The pattern of rental yields falling more in poorer zip codes than richer ones is clearer from the following box plot. Rental yields have fallen across the board. But they have fallen much more dramatically in zip codes with cheaper houses.
We now formally test the hypothesis of differentiated effects by looking at the slope coefficients of changes in log yields against changes in house prices, conditional on house price quartile. In the next graph, the error bars are standard errors of the slope coefficients. We find that the negative gradient is more pronounced for lower quantiles than upper quantiles. This suggests something counterintuitive: rental yields have fallen faster in poorer zip codes with less expensive houses.
Digging deeper, we check whether there is a positive relationship between the level of house prices in 2016 and changes in rental yields since then. We find a clear positive association (t = +18.4), suggesting that, really, rental yields have collapsed more in poorer zip codes than more expensive ones. Why??
This brings us to our second counterintuitive result. The next set of graphs shows the 5-year change in log rent and change in log price, respectively, against log house price level in 2016. We find that rents increased more slowly or have even decreased in more expensive zip codes, while they have increased faster in less expensive zip codes.
Why then have yields fallen more in less expensive zip codes compared to more expensive ones? The answer is that house price appreciation is less pronounced in less expensive zip codes than more expensive ones.
Put another way, while rental yields have fallen across the board because of across the board house price appreciation, they have fallen more in poorer zip codes because house prices have appreciated less in poorer zip codes than in richer ones.
The patterns documented here are suggestive of the influence of the financial cycle. The dramatic increase in house prices and the attendant collapse in rental yields since the pandemic bear the tell-tale signature of the house price channel of monetary policy. The standard 30-year fixed-rate for mortgages have fallen from around 5 percent to 3 percent, while the Case-Shiller National Home Price Index has increased 12 percent over the past year.
Residential fixed investment still has ways to climb, as do home prices. As the mother of all economic booms gets underway, we should expect rental yields to fall even further than they already have. It also has clear implications for investors. The private equity moguls who got in at the bottom are going to make a killing on their housing trade. But it’s still not too late to get in on the trade for other less-than-financially-constrained investors.
Postscript. An earlier version of this article, the one that went out in the email, claimed that rental yields have fallen more in zip codes with cheaper houses relative to zip codes with more expensive houses because rents have increased in the former and fallen in the latter. As I explained on Twitter, this is nonsense. I was confused by a tautology. Both rent and price went up everywhere. Yields fell more in poorer zips relative to richer zips. This can only happen if the rich-less-poor differential in price growth was smaller than rich-less-poor differential in rent growth. Here’s the proof of this lemma:
So, rental yields declined more in poorer places than richer ones because, tautologically, house prices appreciated less there relative to increase in rents. We can think of rental yields as a barometer of the house price cycle. The usual scheme of normalizing by GDP, as we did above for residential fixed investment, is turned upside-down with rental yields: the numerator, rent, plays the role that GDP plays in normalizing investment to extract the cycle.
Interestingly, FT finance editor Robert Armstrong makes the opposite case in his newsletter. Citing no less an authority than Shiller himself, he wagers that the house price boom may be getting out of hand. It’s not an outlandish wager if you believe that inflation is around the corner and the Fed will be forced to hike, thus pushing up mortgage rates and perhaps reducing mortgage credit, thereby leading to a collapse in house prices. But there is no reason yet to believe that inflation has returned. In fact, given the transformation of the inflation process, we should not expect inflation to return any time soon. If my diagnosis of inflation is right, then the housing market has ways to run as the mother of all economic booms gets underway.
Why Have Rental Yields Collapsed?
Great analysis. I do have a question about the last bit, though. Declining rent yields should come from either rents declining or prices increasing, but you say, "Why then have yields fallen more in less expensive zip codes compared to more expensive ones? The answer is that house price appreciation is less pronounced in less expensive zip codes than more expensive ones." Shouldn't that increase rent yields relative to more expensive zip codes? Or am I confused or missing something?
Agree there is money to be made (in any supply-constrained asset, these days). But from a social perspective, isn’t a decline in rental yield simply a transfer from future homebuyers to homeowners. If the trend in rents is steady, nothing has changed from a cashflow perspective - it’s all discount rate. Personally, I think capitalism starts to misfire when society perceives more to gain from falling discount rates than from the growth of underlying cash flows.